Relax retry strategy

[randomir]

In memory/network-starved environment, with (multipart) upload failures occurring, aggressive retry strategy might actually contribute to resource exhaustion.

We should try increasing http_retry_backoff_factor, decreasing number of concurrent upload threads and increasing timeouts, most notably read timeout, as mentioned in https://github.com/dwavesystems/dwave-cloud-client/issues/439#issuecomment-720752001.